Almost every public argument about India’s nuclear liability law is conducted in the language of diplomacy, as if the obstacle to foreign reactors in India were a misunderstanding between governments that better drafting of communiqués could dissolve. It was not. The obstacle was one limb of one section of one statute: clause (b) of Section 17 of the Civil Liability for Nuclear Damage Act, 2010 (Act 38 of 2010), which gave the operator of a nuclear installation a statutory right to recover from a supplier where a nuclear incident had resulted from the supply of equipment or material with patent or latent defects, or from sub-standard services. That clause departed from the international model the rest of the Act faithfully copied, and the departure was deliberate, debated in advance, and consequential in exactly the way its critics in the supplier countries predicted and its defenders in Parliament accepted as a price worth paying. Between the Act’s commencement on 11 November 2011 and its replacement in December 2025, no contract for a foreign supplied reactor was concluded under it.

This article does three things. It states the liability architecture the Act created, precisely, provision by provision, because most accounts paraphrase the decisive clause inaccurately. It gives each side of the controversy its strongest voice and names who held it: the suppliers who refused the exposure, the Government of India which defended the text and then tried to settle its meaning by clarification, rule and insurance pool, the opposition parties which had demanded the clause in the first place, and the critics from the opposite direction who attacked the Act for capping compensation far too low. And it takes a verdict, as this series requires, on the question the record can answer: whether the Act as operated delivered compensation or deterred supply.

Why a liability statute had to exist at all

India operated nuclear power reactors for four decades without any statute governing civil liability for nuclear damage. The Atomic Energy Act, 1962, which is the subject of the article on the atomic energy framework, controlled who could build and run installations and said nothing about who would pay a victim if one failed. That silence was tolerable while every reactor was Indian built, Indian owned and operated by the State, because the practical defendant in any disaster would have been the Government of India itself. It became untenable the moment India sought reactors, fuel and technology from foreign suppliers, because a foreign vendor asked to sell a reactor into a jurisdiction with no liability law faces the worst of all positions: exposure governed by the ordinary law of tort, uncapped, unchannelled and untestable in advance.

The opening came through diplomacy. The civil nuclear cooperation agreement concluded between India and the United States in 2008, and the waiver granted by the Nuclear Suppliers Group in September of that year, ended India’s exclusion from international civil nuclear commerce on the strength of its separation plan and safeguards commitments. What the agreement made possible, it did not itself accomplish. American, French and Russian suppliers each needed a liability regime they could insure against before they could sign supply contracts, and the regime the international industry worked under was a settled one: liability channelled exclusively to the operator of the installation, strict in character, capped in amount, backed by mandatory financial security, adjudicated in a single forum, and extinguished after a defined period. That model, carried in the Vienna and Paris Conventions and generalised by the Convention on Supplementary Compensation for Nuclear Damage adopted in 1997, existed so that a victim would have one solvent defendant, promptly liable without proof of fault, and a supplier could price its risk and buy cover for it.

India signed the Convention on Supplementary Compensation on 27 October 2010, a month after the liability Act received assent, and the Act was designed to make that signature good. Its long title announces the design: an Act to provide for civil liability for nuclear damage and prompt compensation to the victims of a nuclear incident through a no-fault liability regime channelling liability to the operator, with a Claims Commissioner and a Nuclear Damage Claims Commission. The statute’s architecture is the international architecture. Which is what makes the single point of divergence so instructive: Parliament had the model in front of it, followed it in six of its seven elements, and departed from it in the seventh with its eyes open.

Behind the drafting sat a memory no Indian legislator needed explained. The gas leak at Bhopal in December 1984, and the compensation settlement of 470 million United States dollars approved by the Supreme Court in 1989 for a disaster whose medical consequences are contested decades later, had fixed in Indian politics the proposition that a foreign industrial supplier must not be able to walk away from an Indian catastrophe behind a corporate structure. The recourse clause was not an oversight or a drafting accident. It was Bhopal, enacted.

The passage of the Bill: a cap and a clause carried together

The Bill’s parliamentary history explains why the Act took the shape it did, because the two features suppliers and victims’ advocates later attacked from opposite directions, the cap and the recourse clause, were not independent choices. They were a single legislative bargain, and each was the political price of the other.

The Civil Liability for Nuclear Damage Bill was introduced in the Lok Sabha in May 2010 by the government of Prime Minister Manmohan Singh, in a form that tracked the international model closely: liability channelled to the operator, an operator cap of five hundred crore rupees, and a recourse provision in narrower terms. It was referred to the Parliamentary Standing Committee on Science and Technology, which heard the atomic energy establishment, industry, insurers and civil society, and reported in August 2010 with recommendations that pushed the text toward greater victim protection and a higher ceiling. The government accepted a substantial part of that pressure in the official amendments it moved when the Bill returned to the Lok Sabha: the operator’s cap was raised threefold, from five hundred crore to fifteen hundred crore rupees.

The debates in both Houses then ran on two rails at once. Opposition speakers, Sushma Swaraj leading for the Bharatiya Janata Party in the Lok Sabha and Arun Jaitley in the Rajya Sabha, attacked the Bill as drafted to foreign specifications, pointing to the timing of its introduction in advance of an American presidential visit and to the modesty of the sums for which Indian lives were being valued. The Left parties pressed the same attack further and moved amendments to raise the operator’s liability to ten thousand crore rupees and to strip out the ceiling’s protection more broadly; those amendments were negatived. At the same time, and from the same opposition benches, came the insistence that a supplier whose defective equipment caused an incident must not stand immune behind the operator’s cap. The recourse provision emerged from that crossfire in its enacted breadth because it was the one proposition that commanded support on every side of the chamber: the government needed opposition votes or acquiescence for a bill its own diplomacy required, the opposition needed a visible safeguard against the charge that it had let foreign vendors escape, and the clause supplied it. Amendments moved to write the supplier expressly into Section 46 met the opposite fate and were negatived, which is why the saving clause speaks of the operator alone, a point whose forensic significance appears in the Section 46 discussion below.

The Rajya Sabha passed the Bill at the end of August 2010, and presidential assent followed on 21 September 2010. Commencement waited more than a year: the Act and the Civil Liability for Nuclear Damage Rules, 2011 were brought into force together on 11 November 2011 by notification, a delay that itself reflected the government’s awareness that the text it had carried would be read abroad as India’s final word, and that the Rules, which attempted in Rule 24 to bound the recourse the Act had left unbounded, had to be ready to be read with it. The sequence matters for the legal argument that consumed the following decade. Parliament enacted an exposure and fixed no limit to it; the executive, fourteen months later, published rules and then clarifications that purported to supply the limit; and the suppliers’ lawyers advised, correctly as a matter of legal method, that the hierarchy of those instruments runs downward, and that a right created by a section is not cut down by a rule that the section does not mention.

The architecture: who is liable, and on what terms

Section 4 of the Act places liability for nuclear damage caused by a nuclear incident on the operator of the nuclear installation in which the incident occurs. The liability is strict: it arises on proof of the incident and the damage, not on proof of negligence, and the victim is not required to identify which component failed or whose employee erred. The operator under the Act is the Central Government itself, or an authority or corporation established by it, or a Government company holding a licence under the Atomic Energy Act for the installation in question. In practice that meant the Nuclear Power Corporation of India Limited, the public sector operator of India’s power reactors, and it matters to everything that follows that the operator the Act channels liability to is an instrument of the State rather than a private commercial utility.

Who pays first when a nuclear incident causes damage?

The operator of the installation, and only the operator. Section 4 makes the operator strictly liable for nuclear damage caused by a nuclear incident in its installation, without any requirement that a victim prove fault, identify the component that failed, or sue any supplier in the chain behind the plant.

Section 5 then carves out the cases in which even the operator is not liable: where the incident is caused directly by a grave natural disaster of exceptional character, or by an act of armed conflict, hostility, civil war, insurrection or terrorism. Those exclusions track the international model, and they mark the boundary of the no-fault promise. Within the boundary, the scheme is a bargain. The victim surrenders the ordinary law’s wider field of defendants and its uncapped damages, and receives in exchange a single defendant who cannot dispute fault, a dedicated claims machinery instead of a civil trial, and a fund of compensation fixed in advance by statute rather than dependent on the defendant’s solvency at the time of the disaster. Whether that bargain was struck at the right price is the subject of the criticism taken up below; its structure was not in dispute between the parties to the controversy that consumed the Act, because suppliers and the government agreed on the channelling half of the design and fought only over the recourse clause bolted onto it.

What counts as nuclear damage, and how the clock starts

The definition of nuclear damage in Section 2 is wider than the phrase suggests in ordinary speech, and its width matters because the cap, the forum and the limitation period all attach to the defined term. Nuclear damage means loss of life or personal injury, expressly including immediate and long term health impact, and loss of or damage to property, in each case caused by or arising out of a nuclear incident. It extends, to the extent notified by the Central Government, to economic loss arising from that injury or property damage where the person suffering it is entitled to claim; to the costs of measures of reinstatement of an impaired environment, where the impairment is not insignificant and the measures are actually taken or to be taken; to loss of income derived from an economic interest in the use or enjoyment of the environment, lost through a significant impairment of that environment; to the costs of preventive measures, and further loss or damage caused by those measures; and to other economic loss permitted by the general law on civil liability. A fishing community evacuated from a coast, a farmer whose land is taken out of cultivation, a state government that pays for decontamination: each has a head of claim inside the definition, and each claim draws on the same capped fund. The breadth of the definition against the narrowness of the ceiling is the arithmetic the cap’s critics kept returning to.

The process begins before any claim can be filed. Section 3 obliges the Atomic Energy Regulatory Board to notify a nuclear incident within fifteen days of its occurrence, and to give the occurrence wide publicity. A proviso then qualifies the duty: where the Board is satisfied that the gravity of threat and risk involved in an incident is insignificant, it is not required to notify it. The proviso was among the provisions attacked in the Kerala challenge, and the objection to it is structural rather than textual. The regulator whose notification starts the claims process is a body constituted under the atomic energy framework and situated within the same administrative family as the operator it would be notifying an incident against, and the discretion to treat an incident as too insignificant to notify is a discretion to keep the claims machinery from starting at all. No incident tested the proviso during the Act’s life. Its presence in the text still tells a reader something true about the statute’s centre of gravity: wherever the Act had to choose between the programme’s administrative convenience and the claimant’s procedural security, the drafting record is mixed, generous to the claimant on the heads of damage and guarded on the machinery that would convert those heads into payment.

The caps: three figures and a fund

Section 6 fixes the money. Sub-section (1) provides that the maximum amount of liability in respect of each nuclear incident shall be the rupee equivalent of three hundred million Special Drawing Rights, the unit of account of the International Monetary Fund. Sub-section (2) fixes the operator’s own share of that liability, graded by the kind of installation: in respect of nuclear reactors having thermal power equal to or above ten megawatts, rupees one thousand five hundred crores; in respect of spent fuel reprocessing plants, rupees three hundred crores; and in respect of research reactors having thermal power below ten megawatts, fuel cycle facilities other than spent fuel reprocessing plants, and the transportation of nuclear materials, rupees one hundred crores. One crore is ten million rupees, so the operating figure around which the entire controversy revolved was a cap of fifteen billion rupees on the operator of a power reactor for a single incident.

Section 7 then assigns the balance. Where the claims arising from an incident exceed the operator’s capped share, the Central Government is liable for the difference up to the overall ceiling of three hundred million Special Drawing Rights. Where the damage exceeds even that ceiling, the Act contemplates the Central Government taking additional measures to meet it, and once India ratified the Convention on Supplementary Compensation the international fund constituted by that Convention stood behind the national tier as a further source. Section 7 also empowers the Central Government to establish a Nuclear Liability Fund by levying charges on operators, and the Department of Atomic Energy has recorded that the fund was established in December 2015. Section 6 further permits the Central Government to review the operator’s liability figure from time to time and to specify a higher amount by notification. No upward revision of the fifteen hundred crore figure was notified during the Act’s operative life, a point to which the criticism of the cap returns.

Three features of the cap design deserve notice because arguments about the Act habitually miss at least one of them. The ceiling is per incident, not per installation and not per year: a second incident at the same plant in a later year would attract the full statutory structure afresh, which is the design’s one generous dimension. The ceiling binds the compensation payable under the Act; it does not purport to value the damage, and the gap between the two numbers in a severe accident is precisely the unallocated residue that the critics’ case is about. And the rupee figures in Section 6(2) were not indexed. Fifteen hundred crore rupees fixed in 2010 was a smaller sum, in real terms and in dollar terms, with every year the Act lived, so the settlement Parliament struck depreciated steadily without any legislator having to vote for its depreciation. An unrevised, unindexed cap in an inflationary economy is a policy that tightens itself, and the fifteen years over which the figure stood still were years in which the government’s own published answers were obliged to convert the three hundred million Special Drawing Rights ceiling into rupees at a stated rate, three thousand one hundred and sixty-seven crore at the rate the Department of Atomic Energy cited, which is more than double the operator’s tier and a standing illustration of how much of the structure rested on the sovereign rather than on the operator the Act nominally channels liability to.

What financial security did the operator have to carry?

Section 8 required the operator, before beginning operation of an installation, to take out insurance or provide other financial security covering its liability under the Act, for the amount and the period the statute prescribes. That obligation is what the Indian Nuclear Insurance Pool was later created to make performable in the domestic market.

The Indian departure: the Act against the model

The table below is the artifact this article exists to provide. It sets the international liability model, as the Convention on Supplementary Compensation states it, beside the Indian Act element by element. Read down it and the shape of the controversy is visible in a single object: conformity in six rows, one row of divergence, and fifteen years of stalled contracts hanging on that row.

Element of the model	The international convention position	The Civil Liability for Nuclear Damage Act, 2010
Channelling	Liability rests exclusively on the operator of the installation	Liability rests on the operator alone under Section 4; the victim has no claim under the Act against anyone else
Character of liability	Strict liability, without proof of fault	Strict liability under Section 4, subject to the Section 5 exclusions
Cap on liability	Liability limited in amount by law	Overall maximum of the rupee equivalent of 300 million Special Drawing Rights under Section 6(1); operator capped at Rs 1,500 crore for a power reactor under Section 6(2)
Financial security	The operator must maintain insurance or other financial security	Section 8 requires insurance or other financial security covering the operator’s liability
Recourse against suppliers	Available only where a contract in writing so provides, or where the damage was caused by an act done with intent to cause damage	Section 17 allows recourse on those two grounds and adds a third: where the incident resulted from a supplier’s act, including supply of equipment or material with patent or latent defects or sub-standard services. This is the point of divergence
Forum	Claims lie in a single forum, the courts of the installation state, with civil process channelled accordingly	Claims go to a Claims Commissioner, or to a Nuclear Damage Claims Commission where one is established, and Section 35 bars the jurisdiction of civil courts over matters within their remit
Limitation	Rights are extinguished after defined periods from the incident	Section 18 extinguishes the right to claim after ten years for damage to property and twenty years for personal injury, running from the date of the incident

Section 17: the three limbs, stated exactly

Because the controversy turns on the words, the words come first. Section 17 provides that the operator of the nuclear installation, after paying the compensation for nuclear damage in accordance with Section 6, shall have a right of recourse where, first, such right is expressly provided for in a contract in writing; second, where the nuclear incident has resulted as a consequence of an act of the supplier or his employee, which includes supply of equipment or material with patent or latent defects or sub-standard services; and third, where the nuclear incident has resulted from an act of commission or omission of an individual done with the intent to cause nuclear damage.

The first limb is the international model’s own. A supplier and an operator remain free to allocate risk between themselves by contract, and nothing in the conventions prevents a negotiated recourse clause with a negotiated price and a negotiated insurance premium behind it. The third limb is also the model’s: a person who deliberately causes a nuclear incident cannot shelter behind channelling, and no supplier organisation anywhere argued otherwise. The second limb, clause (b), is the Indian addition, and its three features explain the fifteen years that followed.

It is statutory, not contractual. The right exists because the section confers it, whether or not the supply contract says anything about recourse, and therefore whether or not the supplier has priced it, insured it or consented to it in the transaction. A supplier asked to sign a contract silent on recourse was not, as it would have been under the model, protected by the silence.

It is keyed to defect, not to intent or even to pleaded negligence in the ordinary form. The limb is engaged where the incident resulted as a consequence of a supplier’s act which includes the supply of equipment or material with patent or latent defects. A latent defect is by definition one that inspection at delivery does not reveal. Equipment supplied for a reactor is expected to serve for decades, inside systems the supplier does not operate, maintained by people the supplier does not employ, to standards the supplier does not control after handover. A standard that asks, after an incident, whether a defect was latent in material supplied twenty years earlier is a standard whose application no actuary could bound, and the suppliers said so in almost exactly those terms.

And it is not capped by the section itself. Section 17 states the right without stating its measure. The operator’s recovery could be argued up to the compensation the operator had paid, and the section fixes no period after supply within which the exposure ends. The contrast with the rest of the Act is stark: the victim’s claim is capped, channelled, forum-bound and time-barred by the statute, while the operator’s claim against the supplier, under clause (b), was none of those things on the face of the section. The attempts to bound it came later, by rule and by executive clarification, and the legal weakness of those instruments is a large part of the story.

Clause (b) also leaves its own working questions unanswered on its face, and each unanswered question was a line item in a supplier’s risk assessment. The limb requires that the incident has resulted as a consequence of the supplier’s act. Causation in a reactor incident is rarely singular: a component may be defectively made, defectively specified by the buyer’s own engineers, defectively installed by a contractor, and defectively maintained for a decade, and an incident may require all four. The section gives no rule for apportionment among causes, no standard by which the supplier’s contribution is to be weighed against the operator’s operation of the plant after handover, and no indication whether the operator must prove the defect caused the incident or merely contributed to it. The phrase patent or latent defects borrows the vocabulary of sale of goods law, where it operates between a buyer and a seller in respect of goods the buyer can reject, return or claim the price of; transplanted into a compensation statute, it has to bear the weight of causation for a radiological event, a task for which the borrowed vocabulary was never designed and on which no Indian court has pronounced, because no recourse action under the clause was ever brought. The services limb is wider still: sub-standard services is a standard without a stated benchmark, in an industry where the relevant services, design review, commissioning support, training, maintenance advice, run across decades and across the boundary between the supplier’s personnel and the operator’s. A supplier reading the clause in 2011 could not tell its board what conduct would keep it safe, because compliance with the contract, with the regulator’s requirements and with international practice were none of them stated to be a defence. The only certain route to safety under clause (b) was not to supply. That is not a rhetorical flourish; it is the arithmetic the vendors’ advisers presented, and the fifteen-year record is consistent with boards having accepted it.

There is also the question of who within the supplier the clause reaches. Its text runs against the supplier, and the recourse is a money claim; but an incident investigation that fixes on a latent defect fixes on people as well as on a company, and Section 46’s preservation of other proceedings, considered next, kept alive the possibility that the same facts would ground regulatory or criminal process against individuals. The Bhopal prosecutions, in which the chairman of the parent company became the most famous absconding accused in Indian industrial history, were part of every foreign board’s briefing on India. Clause (b) was debated as a compensation provision. It was assessed, in the rooms where the decisions were taken, as one element of a total exposure that included reputational and personal dimensions no insurance pool could indemnify.

Section 46: the clause that names the operator and the fight over the missing words

Section 46 is short, and its brevity is the second half of the controversy. It provides that the provisions of the Act shall be in addition to, and not in derogation of, any other law for the time being in force, and that nothing contained in the Act shall exempt the operator from any proceedings which might, apart from this Act, be instituted against such operator.

Read with the channelling design, the clause has an innocent purpose. The Act is a compensation statute, not a general immunity. It does not touch criminal liability under the penal law, regulatory proceedings under the atomic energy framework, or environmental proceedings against an operator who has polluted, and Section 46 says so. A saving clause of this kind is ordinary drafting.

The suppliers read it differently, and their reading was not paranoid. If the Act is in addition to and not in derogation of any other law, and if nothing in it forecloses proceedings that might otherwise be instituted, then the question is against whom those other proceedings might be instituted. The clause expressly preserves proceedings against the operator. It does not expressly preserve them against a supplier, but neither does it expressly close them, and the Act nowhere states in terms that no proceeding of any kind shall lie against a supplier for nuclear damage. A victim, or a victims’ organisation, could therefore be advised to sue a supplier in tort under the general law, outside the Act, unconstrained by the Act’s cap, its forum and its limitation periods, and Section 46 would be the first paragraph of the plaint. The suppliers’ position, stated by the American, French and Russian vendor organisations in their engagement with the Government of India and repeated in the legal opinions their counsel obtained, was that the combination of clause (b) of Section 17 and the open texture of Section 46 left them exposed through two doors: recourse by the operator after payment, and direct suit by victims outside the Act altogether.

The textual point about the missing words is strengthened by the parliamentary record. During the passage of the Bill, amendments were moved to Section 46 which would have extended its terms expressly to the supplier of material, design or services, preserving proceedings against such a person in Indian or external courts. Those amendments were put to the vote and negatived. The Government of India later relied on exactly that history: Parliament had considered writing the supplier into Section 46, had declined, and the enacted clause therefore speaks only of the operator. The suppliers’ lawyers answered that a negatived amendment is a frail foundation for a multi-billion dollar exposure decision, because courts construe enacted words, and the enacted words preserve other law generally while naming only the operator for a distinct proposition. Both readings were honestly available. That is precisely why the clause frightened a risk committee: the dispute was not about what the Government of India intended, but about what an Indian court, seised of a disaster and a sympathetic plaintiff, might one day hold the words to mean.

The doctrinal argument about Section 46 ran deeper than the drafting history, and it is worth stating at the level the lawyers argued it, because it shows why neither side could be given the assurance it needed. The government’s construction, developed in the 2015 answers and defended by commentators sympathetic to it, including Dr M.P. Ram Mohan of the Nuclear Law Association, rests on the character of the Act as a complete and special code for nuclear damage. On that reading, a claim for nuclear damage exists only under the Act: the Act creates the right, channels it to the operator, fixes its forum and its ceiling, and Section 46 cannot resurrect against a supplier a claim that the code has abolished by omission, because the proceeding the clause preserves are proceedings the general law otherwise allows, and the general law of tort cannot supply a claim for a species of damage that a special statute has comprehensively redefined. The contrary construction, pressed in the suppliers’ advice and in critical commentary, starts from the clause’s own words. Parliament did not enact that no claim for nuclear damage shall lie except under this Act. It enacted a schedule of liability for the operator and then added that the Act is in addition to and not in derogation of any other law. A court faithful to the enacted sequence, the argument runs, will not read a code into existence out of a structure, against an express saving of other law, in a case where the plaintiff is a radiation victim and the defendant a foreign manufacturer: the interpretive presumptions in such a courtroom do not run in the supplier’s favour, and everyone advising on the question knew it. Both constructions are competent. That is the point. A supplier deciding whether to commit a reactor design, a supply chain and a balance sheet to India for sixty years was being asked to bet that the first construction would prevail in litigation that had not happened, before courts that had not been seised, in the aftermath of an accident whose facts would dominate any legal analysis offered to them. Boards do not take that bet when other markets exist, and the legal profession’s inability to close the question, rather than any government’s bad faith, is the technical core of the fifteen years.

The claims machinery: a commissioner instead of a court

The compensation half of the Act deserves the same precision, because the critics who attacked it from the victims’ side were attacking real provisions, not the controversy the diplomats were having.

An application for compensation under the Act may be made by the person who has sustained injury, by the owner of property that has suffered damage, by the legal representatives of a person whose death has resulted, or by an agent authorised by any of them. The Atomic Energy Regulatory Board must first notify the nuclear incident, and the claims process runs from that notification. Applications go to a Claims Commissioner appointed by the Central Government for the area in which the incident occurred. The Commissioner inquires into the claims, following a summary procedure rather than the full apparatus of a civil trial, and makes an award. Where the Central Government considers that the scale of damage calls for it, the Act provides instead for a Nuclear Damage Claims Commission, established by notification, with a chairperson and members, to discharge the same adjudicatory function across a wider field. Section 35 bars the jurisdiction of civil courts in respect of matters which the Commissioner or the Commission is empowered to determine, so the statutory forum is not an alternative to the courts but a replacement for them. This account states the law as enacted and as later replaced; it is legal information about a repealed statute, not advice on any individual claim.

Section 18 fixes the outer life of a claim. The right to claim compensation is extinguished if the claim is not made within ten years from the date of the nuclear incident in the case of damage to property, and within twenty years from the date of the incident in the case of personal injury to any person. Those periods are long by the standards of ordinary tort limitation, and they were set with radiation injury in mind, but they run from the incident and not from the discovery of the harm, and a radiation-induced disease that declares itself in the twenty-first year falls outside the Act altogether. The latency objection, pressed by public health campaigners and by the petitioners who later challenged the Act, is that a limitation clock which starts before the injury exists will extinguish claims that no diligence could have brought in time. The government’s answer, in substance, was that some terminal date is indispensable if liability is to be insurable at all, that twenty years sits within the range the international instruments contemplate, and that a claim extinguished under the Act is a claim against a compensation fund that was never designed to stand open for a lifetime. That answer is coherent as insurance logic. As an answer to a widow whose husband’s cancer is diagnosed in year twenty-one, it is the weakest joint in the compensation half of the statute.

The choice of a commissioner over a court was itself a considered feature of the design, and its reasoning should be recorded before it is criticised. India’s experience of mass disaster litigation was Bhopal, where claims ground through a settlement, a claims scheme and decades of review petitions, and where the interval between the disaster and anything resembling final payment was measured in the lifetimes of the injured. A dedicated commissioner, deciding on a summary procedure in the affected area, against a fund that exists in advance and a defendant who cannot contest fault, was designed to invert every term of that experience: local rather than distant, administrative rather than forensic, funded rather than contingent. The Commission variant extends the same logic to scale, substituting a multi-member body where the damage exceeds one officer’s capacity. The costs of the design are the mirror of its virtues. Summary procedure compresses the testing of evidence in the class of case where evidence is most technical; a single channelled forum removes the possibility of a test case developing the law; and the bar on civil courts means that the quality of justice under the Act is exactly the quality of the appointments made under it, with no alternative route when appointments lag. No appointment was ever tested, because no incident occurred, and the machinery section of the Act therefore stands in the record as pure design: never operated, never reformed, and since repealed, carrying its promise of prompt payment away with it unexamined. A reader weighing the Act should hold that fact beside the supplier story. The statute’s famous half, the recourse clause, was fought over for fifteen years by the most powerful governments on earth. Its other half, the half written for the victims, was never used at all, and nobody with power treated that as an emergency.

How other regimes answer the same questions

India’s choices stand out more clearly against the regimes its suppliers worked under elsewhere, because the divergences were specific rather than general, and a supplier comparing jurisdictions was comparing clauses.

The United States’ Price-Anderson Act, the oldest of the modern regimes, channels liability to the operator as a matter of economic reality rather than by prohibiting suit: claims arising from a nuclear incident are consolidated in the federal courts under a public liability action, suppliers are indemnified through the operator’s cover, and the fund is built from operator insurance topped by retrospective premiums levied across the industry, producing an aggregate measured in billions of dollars per incident. The American vendor that declined Indian exposure was not refusing liability law as such; it was refusing a regime in which it could be pursued outside the channel, in a local forum, without the indemnity structure its home statute gave it.

France, whose vendor negotiated for Jaitapur, works under the Paris Convention with operator liability, caps revised upward over time, and recourse confined on the conventional pattern. Russia, whose vendor built at Kudankulam, is a party to the Vienna Convention and in any event supplied India as a state under intergovernmental agreements. Japan, which concluded a civil nuclear cooperation agreement with India in 2016, imposes operator liability without a statutory ceiling of the Indian kind, supported by government indemnity arrangements, and its experience at Fukushima demonstrated the fiscal reality behind every regime in this field: where damage reaches a catastrophic scale, the state pays, whatever the statute says about caps, because no private balance sheet and no insurance market carries a catastrophe of that order. Germany’s atomic law likewise works with unlimited operator liability backed by financial security requirements.

Two conclusions follow from the comparison, and they cut in opposite directions, which is why the comparison is worth making carefully. In the suppliers’ favour: India’s combination of a low operator cap, a statutory defect-based recourse without ceiling or sunset, and a saving clause that did not close the tort door was, taken together, outside the range of regimes any major vendor had accepted anywhere, and the vendors’ refusal was consistent with their conduct in every other market. In the critics’ favour: the regimes held up as models are not obviously more generous to victims. Price-Anderson’s channel is as exclusive as anything in the Indian Act, and the international conventions’ caps are set, like India’s, by reference to what can be insured. The honest comparative statement is that India’s Act was orthodox in the restrictions it placed on victims and heterodox only in the exposure it preserved against suppliers, and that the world’s vendors had organised their affairs around the orthodoxy in both directions, accepting capped and channelled victim claims as the price of a regime in which their own position was equally capped and channelled. Clause (b) broke the symmetry on one side only, and it was the broken symmetry, more than the size of any figure, that the market refused.

The suppliers’ case, at full strength

Here is the case against clause (b), made as the suppliers and their governments made it, without interruption.

Channelling is not a favour to suppliers; it is the mechanism that makes victims’ compensation payable. The international regime concentrates liability on the operator for three connected reasons. A victim should not have to identify which of dozens of contractors across several countries supplied the component that failed. The operator, who runs the plant, controls the site and answers to the regulator, is the party whose conduct the liability signal should reach. And a single liable party with mandatory insurance produces a fund that exists on the day of the disaster, rather than a decade of litigation against a supply chain. A statute that reopens the supply chain after the operator has paid does not merely burden suppliers; it reintroduces, one step removed, the fragmentation the regime exists to prevent, because the operator’s recourse action will be litigated exactly as a supply chain dispute, with the operator’s payment to victims hostage to its outcome in every case where the operator’s own balance sheet cannot carry both.

The exposure clause (b) creates is not priceable. Commercial supply requires a vendor to quantify its worst case, insure it, and load the premium into the bid. Clause (b) defeats each step. Its trigger, a latent defect in equipment or material, can be alleged in respect of any component however old, however maintained, however operated after delivery. It has no statutory ceiling of its own and no statutory sunset. The defect standard invites a retrospective engineering inquest conducted under litigation pressure after a catastrophe, in which the question will not be what a reasonable supplier should have foreseen but what a tribunal, knowing the outcome, can be persuaded was latently wrong. No nuclear insurer in the world wrote cover for that exposure on ordinary terms, and the vendors’ home governments, the United States most visibly, made conformity with the Convention on Supplementary Compensation a condition of their own support for projects, since the Convention’s annex requires national law to channel liability to the operator with recourse confined to contract and intent.

Section 46 compounds the defect rather than causing it, because it leaves the tort door unlatched. A supplier which accepted clause (b) as a bounded commercial risk would still face the possibility that victims sue it directly under other law, outside the cap. The two exposures together meant that signing an Indian supply contract was not a decision a vendor’s board could take on legal advice that the risk was closed; the advice available was that the risk was arguable, which for a board is a different thing.

And the predicted consequence was not a richer price but no contract. Nuclear vendors are few, their order books are global, and India was one market among several competing for the same engineering capacity. A jurisdiction whose liability terms cannot be insured is not a jurisdiction that pays more for reactors; it is a jurisdiction that is not supplied. That prediction was made in 2010 and 2011 by named vendor organisations and by the governments of the United States and France on their behalf, and the fifteen years that followed are the evidence for it: the sites were allotted, the intergovernmental agreements were signed, the techno-commercial negotiations ran on, and no foreign reactor supply contract under the Act was concluded.

One further element of the suppliers’ case deserves record because it explains why the insurance pool, however well designed, could not by itself close the argument. Nuclear liability insurance is a specialist global market, written through national pools that reinsure among themselves, with the London market’s nuclear risk insurers standing behind much of the world’s capacity. A pool can only sell a supplier a policy whose trigger it can define, and the trigger Section 46 left open, a direct claim under the general law outside the Act, was not a risk the Indian pool’s policy was written to answer, because it was not a risk under the Act at all. The suppliers’ recourse policy therefore covered the symmetrical, bounded part of the exposure and necessarily stopped where the statute’s asymmetry began. Vendor boards were not refusing insurance; they were refusing a residue that no insurer on earth had offered to carry, and the distinction matters to the verdict, because it converts the suppliers’ conduct from a negotiating posture into a market fact. A posture can be waited out or bought off. A gap in the world’s insurance capacity cannot, and the governments that pressed India for amendment through five years of clarification were transmitting, in diplomatic language, the conclusion of underwriters who do not attend summits.

The Indian suppliers inside the same net

The controversy is usually narrated as a dispute between the Indian state and foreign vendors, and that narration conceals a domestic fact the briefs of the period were careful about: Section 17 draws no line between a foreign reactor vendor and an Indian engineering firm. The term supplier, as the 2011 Rules elaborated it, reaches a person who manufactures and supplies equipment or material, who provides a design or a service, or who stands in the chain of supply to the operator, and the recourse right in clause (b) runs against any of them whose act, including the supply of defective equipment or sub-standard services, results in an incident.

India’s indigenous programme had built a substantial domestic supply chain over decades, heavy engineering, forgings, turbines, instrumentation and construction in the hands of Indian public and private companies, and those companies read Section 17 as addressed to themselves as much as to any American or French vendor. Their exposure differed from the foreign vendors’ in one respect only: they had no alternative market to walk to, and no home government to negotiate for them. Indian industry’s representations during the Act’s early years sought the same two things the foreign vendors sought, a ceiling and a clock on recourse, and the Nuclear Power Corporation’s own contracting practice shows how the system adapted where it could. The Corporation’s purchase documents for supply contracts incorporated the Section 17(a) route expressly, writing the recourse into the contract on the terms Rule 24 prescribed, recourse limited to the lower of the operator’s liability and the contract value, for the duration of the initial licence or the product liability period, whichever was longer, and referring questions on supplier liability to the government’s clarifications of February 2015. In its domestic contracting, in other words, the operator itself treated the contractual limb, bounded by rule, as the workable instrument, and treated clause (b) as a background presence rather than the operative term. That practice is evidence of a kind rarely available in statutory interpretation disputes: the party the clause was written to protect conducted its ordinary business as if the bounded contractual version of recourse were the real one, and reserved the statutory version, unexercised and untested, for the catastrophe that never came.

The domestic dimension also complicates the morality play in both directions. Against the suppliers’ case: the plea that clause (b) made supply impossible sits oddly beside a domestic chain that continued to supply under the same Act, albeit with recourse written into its contracts on bounded terms. Against the clause’s defenders: the domestic chain’s acceptance was the acceptance of firms dependent on a single state buyer, and their quiet absorption of the risk is evidence of their bargaining position, not of the risk’s reasonableness. What the domestic record establishes without strain is narrower and more useful: where recourse was contractual, capped by reference and limited in time, supply continued; where it was statutory, uncapped in the section and unlimited in time, the counterparties with a choice declined. The variable that predicted behaviour was not nationality. It was boundedness.

The case for the clause, and the government’s long attempt to settle its own statute

The case for clause (b) was made across the political spectrum during the Bill’s passage, and it is not a weak case. The government of Prime Minister Manmohan Singh introduced the Bill in a form closer to the international model; the recourse provision in its enacted breadth, and the resistance to raising the cap and to softening Section 46, came under sustained pressure in Parliament from the Bharatiya Janata Party and from the Left parties, whose speakers made three arguments that any honest account must state at full strength.

A supplier of defective equipment should not be immune. If a valve, a vessel or a control system fails because it was defectively made, a legal regime that forbids the paying party from recovering against the maker does not merely allocate a loss; it extinguishes a deterrent. The threat of recourse disciplines design review, manufacturing quality and disclosure of known weaknesses in a way that a regulator’s inspection regime, however diligent, cannot replicate, because the supplier knows its own product’s weaknesses in a detail no inspector will match. India’s industrial history supplies the emotional core of the argument: at Bhopal, the corporate structure and the settlement together meant that the entity whose plant killed thousands paid a sum fixed by negotiation, and the political class that watched that outcome was determined that a nuclear supplier would not enjoy a statutory promise of immunity in advance.

Exclusive channelling was designed for a different market structure. In the countries whose practice formed the model, the operator is typically a private or corporatised utility which buys the plant, prices the liability risk into its tariff, and stands between the supplier and the public. In India the operator is the State in corporate dress. Exclusive channelling in India therefore does not place the cost of a supplier’s defective equipment on a commercial intermediary able to negotiate it away in the purchase price; it places the cost on the Consolidated Fund, which is to say on the taxpayer, including the taxpayer who lives beside the plant. A Parliament asked to enact that the Indian public shall indemnify a foreign vendor against the consequences of the vendor’s own defective supply was entitled to refuse, and the refusal was expressed by parties that agreed on little else.

The cap and the recourse clause must also be read together politically. The same Parliament that was asked to limit victims’ recovery to a fixed sum was being asked to immunise suppliers entirely. The legislative bargain that emerged, a cap for victims paired with a recourse right for the operator, distributed the statute’s severities: the victim accepted a ceiling, and the supplier accepted an exposure. To describe clause (b) as an aberration is accurate against the international model and incomplete against the domestic politics, in which the clause was the condition on which the cap itself was carried.

Rules, answers and a pool: the executive settlement of 2011 to 2016

What no defender of the clause could deny was that the statute, as enacted, left the exposure unbounded in amount and time. The executive branch of two successive governments then spent five years trying to bound it without amending it, and the instruments it used, and their legal limits, are the second act of the controversy.

The first instrument was subordinate legislation. The Civil Liability for Nuclear Damage Rules, 2011, made when the Act was brought into force, addressed recourse in Rule 24. A contract of the kind described in clause (a) of Section 17 was required to include a right of recourse of not less than the extent of the operator’s liability under Section 6(2) or the value of the contract, whichever is less, which in operation set the contract’s recourse measure at the lower of the operator’s statutory liability and the contract price. And the recourse provision was to run for the duration of the initial licence issued under the Atomic Energy (Radiation Protection) Rules, 2004, an initial licence period of five years, or the product liability period for which the supplier has undertaken liability for patent or latent defects or sub-standard services under the contract, whichever is longer. Rule 24 thus supplied the two things Section 17 omitted: a ceiling by reference and a clock.

Its limits were pointed out at once. The rule by its terms governs contracts under clause (a). It does not in terms reach clause (b), which is a statutory right independent of any contract, and a supplier reading the Act and the Rules together could not be sure that a court would import the rule’s ceiling and clock into a clause (b) action. Soli Sorabjee, the former Attorney General, gave a written opinion in the period of the Rules’ making that Rule 24 was arguably inconsistent with the Act it purported to implement and vulnerable as beyond the rule-making power, on the ground that a rule cannot scale down a liability the parent statute has imposed in unqualified terms. The government’s lawyers took the opposite view of their own rule. A supplier’s board, presented with a rule whose validity was contested by one of India’s most eminent counsel and whose reach did not plainly cover the contested clause, had not been given certainty; it had been given a better argument.

The second instrument was clarification. On 8 February 2015 the Ministry of External Affairs published a set of frequently asked questions and answers on the Act and related issues, the question-and-answer document circulated in office formats and readable through an office file viewer for Excel, Word and PowerPoint documents, in which the government’s settled position was stated in terms. On Section 17, the answers said that the provision permits but does not oblige the operator to exercise recourse, that recourse under clause (a) operates where the contract so provides and within the contract’s own limits, and that the operator’s discretion in structuring its contracts was the mechanism the Act itself supplied. On Section 46, the answers said that the provision does not create any liability for a supplier and does not override the Act’s channelling, since Section 46 speaks of proceedings against the operator and the Act’s scheme confines claims for nuclear damage to the statutory forum. On the Convention on Supplementary Compensation, the answers maintained that Indian law was consistent with the Convention, whose annex, the government argued, permits recourse in the circumstances the Act provides. The external affairs ministry described the ratification that followed as a conclusive step in addressing the liability issues.

The criticism of the clarification was made immediately and has never been answered on its own ground: an executive answer sheet cannot amend, narrow or gloss an enacted provision against its text. If Section 17(b) means what it says, an answer saying that the operator is free not to use it addresses the operator’s discretion, not the supplier’s exposure once a contract is signed and an incident has occurred, because the discretion belongs to a public corporation answerable for public money, whose decision not to pursue a statutory recourse after a disaster would itself be politically and legally contestable. If Section 46 leaves a tort claim open, an answer saying that it does not is an assertion a future court is free to disregard. The clarification bound nobody who would decide the question. It was, in form, a diplomatic instrument: its real audience was the United States administration, which had pressed the liability question at summit level, and its real function was to record the Indian state’s interpretation in a form foreign governments could rely on politically even where their vendors’ lawyers could not rely on it legally.

The clarifications did not arrive in a vacuum, and their diplomatic setting explains their form. In December 2014 the two governments constituted a contact group on civil nuclear cooperation, charged with resolving the outstanding issues between the Indian system and its American counterparties, liability at their head. When President Barack Obama visited New Delhi in January 2015, the two governments announced a breakthrough understanding on civil nuclear cooperation, and the liability limb of that understanding consisted of precisely the instruments described here: the Indian interpretation to be recorded in the answers document, the insurance pool to be constituted, and ratification of the Convention to follow. The sequence is worth noticing for what it reveals about where the dispute was actually settled and where it was not. At the level of governments, the understanding held: the answers were published in February 2015, the pool was launched in June 2015, ratification followed in February 2016, and official statements on both sides treated the liability question as addressed. At the level of the counterparties who would have to sign, insure and build, nothing in the sequence altered the text of Section 17 or Section 46, and the contracts the understanding was meant to unlock did not follow. A settlement that satisfies the principals’ political requirements while leaving the agents’ legal requirements unmet is a settlement of the communiqué, and the decade after 2015 is the record of that distinction working itself out.

The third instrument was insurance. On 12 June 2015 the General Insurance Corporation of India, GIC Re, together with eleven other Indian insurers, launched the Indian Nuclear Insurance Pool with a capacity of fifteen hundred crore rupees, matching the operator’s statutory cap. The pool offered two covers: the operator’s liability policy, insuring the Nuclear Power Corporation against its Section 6 liability, and a suppliers’ right of recourse policy, under which a supplier could insure against the very exposure Section 17 created, with the New India Assurance Company issuing policies on the pool’s behalf. The pool’s architects, led by GIC Re under its chairman and managing director Alice Vaidyan, presented it as the twenty-seventh nuclear insurance pool in the world and as the practical answer to the objection that the exposure was uninsurable: the exposure had been given a price, and a supplier could buy it. The suppliers’ response exposed the remaining difficulty. A recourse policy insures the supplier against the operator’s claim; it does not, and could not, insure against a direct tort claim under Section 46 brought outside the Act, and the premium prices a risk whose legal boundaries were still the contested ones. The pool solved the Section 8 problem completely and the Section 17 problem partially, and it could not reach Section 46 at all. The arithmetic of the settlement is worth stating plainly. A pool of fifteen hundred crore rupees equals the operator’s statutory cap, so the operator’s tier is, on paper, fully covered against one incident; a severe incident, however, draws on the government’s tier up to the 300 million SDR ceiling and beyond it on the government’s general powers and the Convention fund, and none of those larger resources passes through the pool. The Nuclear Liability Fund, levied from operators from December 2015, was the intended long-run source of the government’s tier, but its collections, like the pool’s capacity, were built for the incidents the Act’s economics assumed, not for the incidents its critics invoked. The settlement package, in other words, perfected the Act’s intended design and left the design’s central gamble, that the capped sums would suffice, untouched. Its designers understood the bargain. A pool administrator pricing cover for a catastrophe the fund could not absorb is not providing against catastrophe; it is providing against the political appearance of unpreparedness, which is a different service and was, in the context of the supplier negotiations, the service being bought.

The fourth instrument was ratification. India deposited its instrument of ratification of the Convention on Supplementary Compensation with the International Atomic Energy Agency on 4 February 2016, and the Convention entered into force for India on 4 May 2016. Ratification mattered in two ways. It gave India access to the Convention’s international tier of compensation beyond the national ceiling. And it was a public assertion by the Indian state that its law conformed to the Convention, an assertion the United States, itself a party, was prepared at government level to accept, while France, a Paris Convention state, and Russia, a Vienna Convention state, stood outside the Convention altogether and made their own assessments. The diplomatic track had, by 2016, delivered everything diplomacy could deliver: a clarification, a pool and a ratification. What it had not delivered was a signature on a reactor supply contract, and the gap between those two lists is the measure of the difference between settling governments and settling a text.

What did the Nuclear Liability Fund add?

Section 7 empowered the Central Government to build a Nuclear Liability Fund out of levies on operators, so that the state’s tier of an incident’s cost would rest on a collected fund rather than on a budgetary scramble after the event. The Department of Atomic Energy records the fund as established in December 2015.

The critics from the other direction: a cap too low, a forum too narrow, a clock too short

The suppliers’ controversy consumed the Act’s public life so completely that the second critique, aimed at the compensation half of the statute, received a fraction of the attention, although on its own terms it is at least as substantial. It was pressed by victims’ advocates, by public health campaigners in the Bhopal movement’s tradition, by opposition parliamentarians during the Bill’s passage who demanded a far higher ceiling, and by the petitioners who carried the Act to court.

The cap is the centre of it. Fifteen hundred crore rupees on the operator, and an overall statutory maximum of the rupee equivalent of three hundred million Special Drawing Rights, which the Department of Atomic Energy’s own published answers valued at roughly three thousand one hundred and sixty-seven crore rupees at the rate they cited, are sums fixed by reference to insurability and to the Convention’s floor, not by reference to the cost of a serious nuclear accident. The comparisons the critics made were not exotic. The Bhopal settlement of 470 million dollars was reached for a disaster at a chemical plant in 1989, and the medical and environmental costs of that disaster continue to be argued over. The costs of the Fukushima accident in Japan, in compensation, decontamination and decommissioning, exceeded India’s entire statutory maximum by orders of magnitude within a few years of the event. A Parliament that fixed the ceiling in 2010 and never revised it, despite holding an express power in Section 6 to raise the operator’s figure by notification, made a choice that deserves to be examined as a choice: the ceiling protected the programme’s economics, and the risk of the excess was transferred silently to the population around the plants, who are the one party to the entire arrangement that was never at any negotiating table.

The public money point compounds it. The operator is a government company; its cap is insured through a pool of predominantly public insurers; the tier above the cap is the Central Government’s; and the fund is levied within a state-owned system. At every level, the money that answers for a nuclear incident in India is public money. The channelling model, designed to interpose a commercial operator between the supplier and the victim, in India interposes the exchequer, and the victim’s compensation and the supplier’s indemnity are drawn from the same pocket. Critics who supported the recourse clause made exactly this point in its favour, and critics who attacked the cap made it against the Act’s distribution of sacrifice: the statute was generous in the certainty it gave the programme and exact in the limits it set on the injured.

The forum and the clock complete the critique. Section 35 removes the civil courts, whose procedures, for all their delays, carry rights of evidence, cross-examination and appeal developed for contested causation, and substitutes a commissioner deciding by summary process, in a field where the decisive questions, which exposure caused which cancer, are the most technically contested questions in all of tort law. The Supreme Court’s own environmental jurisprudence, from the oleum gas leak case onward, built an absolute liability doctrine for hazardous industry precisely on the ground that enterprises profiting from dangerous activity must answer for its harms without the shelter of fault defences, a doctrine traced in the guide to the Environment (Protection) Act, 1986; a statute that caps the answer and removes the courtroom sits uneasily beside that doctrine, and the unease is principled rather than rhetorical. The industrial safety regime whose failures framed Bhopal is the province of the article on the Factories Act, 1948, and the comparison is instructive: factory law at least left the injured their ordinary remedies alongside its inspectorate. The liability Act took the ordinary remedies away and replaced them with a schedule.

The regulatory setting deepened the critics’ case in a way the liability debate rarely pauses on. The Atomic Energy Regulatory Board, the body charged with nuclear safety oversight and with notifying the incident that starts the claims process, owed its existence to an executive notification under the Atomic Energy Act rather than to a statute of its own, and it sat within the atomic energy establishment whose promotional and operating arms it was meant to police. The Comptroller and Auditor General and parliamentary committees had both remarked on the anomaly before the Act, and the government introduced a Nuclear Safety Regulatory Authority Bill in 2011 to place the regulator on a statutory and independent footing; the Bill was never enacted, and it lapsed. For the critics of the liability Act, the two statutes’ fates belonged in one sentence: Parliament in 2010 built a compensation code that trusted the establishment to notify incidents, cap claims and staff the forum, while the companion reform that would have made the establishment’s safety arm independent of it was allowed to die. A claimant under the Act therefore faced a structure in which the operator, the incident notifier and the claims apparatus all drew authority from the same department of government, and the absence of any incident during the Act’s life is the only reason that structure was never stress-tested in public. The SHANTI framework’s grant of a statutory footing to the regulator, a decade and a half later, is an implicit concession that the critics of 2010 had identified a real defect, and it arrived, characteristically, as part of the package whose selling point was not victim protection but investment.

The Convention mechanism in operation

The Convention on Supplementary Compensation deserves a section of its own because it is the standard against which the Indian Act was measured, and the measurements came out differently in government offices, in suppliers’ memoranda and, most importantly, in the text itself. The Convention, adopted in 1997 under the International Atomic Energy Agency, organises compensation in tiers. The installation state must ensure the availability of at least three hundred million Special Drawing Rights for a nuclear incident, by the channelled operator liability, by public funds, or by a combination of the two. Above that tier stands an international fund, to which contracting parties contribute according to a formula based on their installed nuclear capacity and their United Nations rate of assessment, disbursed when the damage of an incident exceeds the national tier. The design’s purpose is to add solidarity where national resources end, without requiring any state to price the unbounded.

The Convention’s entry into force itself illustrates how nuclear liability law moves at the pace of energy politics. It needed five ratifying states with at least four hundred thousand units of installed nuclear capacity, a threshold that kept it dormant for years after signature; it was Japan’s acceptance in January 2015 that finally brought it into force that April. India’s ratification in February 2016 placed it inside a club whose other members included the United States, Japan, Argentina, Romania, Morocco, Montenegro and the United Arab Emirates, conspicuously excluding France and Russia, which remained within their own convention traditions, the Paris and Vienna systems respectively. That asymmetry mattered to the negotiation. For the United States, the Convention’s annex, which requires the installation state’s law to channel liability to the operator and confines recourse to contract or intent, was the legal hook of its objection to clause (b). For France and Russia, parties to the older conventions rather than the supplementary one, the objection was the same in substance but carried without the annex, as a proposition about the global regime’s common understanding. India’s position, pressed in the 2015 answers and in every bilateral engagement after, was that clause (b) was a form of recourse the annex could accommodate, since the annex permits recourse in accordance with national law’s contract and intent grounds and, on India’s reading, did not exhaustively forbid others. The honest description of the disagreement is that the annex’s text was written to describe the existing conventions’ practice, the Indian Act claimed to fit within it by construction, and no authoritative interpreter of the Convention ever ruled on the fit, because the Convention has no court and its secretariat does not issue opinions on members’ statutes. The conformity question was therefore permanently arbitrable and never arbitrated, which is why both governments could simultaneously assert their positions in good faith for a decade and a half.

The constitutional challenge: examined in one court, upheld in another

The Act was challenged on precisely the grounds the second critique identifies. A public interest petition filed by the non-governmental organisation Common Cause, argued by advocate Prashant Bhushan, attacked the limitation of liability and the supplier-protective features of the statute as violations of the right to life, on the reasoning that a polluter-pays and absolute liability jurisprudence forbids Parliament to cap the consequences of a hazardous enterprise and indemnify its participants in advance. In March 2012 a Supreme Court bench headed by Chief Justice S.H. Kapadia agreed to examine the constitutional validity of the Act, observing that the statute had a link to the right to life, while declining the petition’s wider demands for safety audits and an independent regulator on the ground that the Court lacked the scientific competence to run a reactor programme. The examination the bench promised did not produce a concluding judgment on the Act’s validity during the statute’s operative life; the petitions outlived the framework they attacked.

The challenge that did reach a reasoned conclusion was heard in Kerala. A public interest petition before the Kerala High Court assailed a long list of provisions, including Sections 4, 5, 6, the appointment provisions, Section 18’s limitation for personal injury, Section 19, Section 35’s bar on civil courts and the Commission provisions, as violating Part III of the Constitution. A Division Bench of Chief Justice Ashok Bhushan and Justice A.M. Shaffique dismissed the petition, recording that the bench was not satisfied that the enactment suffered from any infirmity, arbitrariness or violation of fundamental rights. Dr M.P. Ram Mohan, president of the Nuclear Law Association, welcomed the dismissal on two grounds: that it reinforced India’s case that the Act was compatible with the international regime, and that it held the statute consistent with the Supreme Court’s absolute liability decisions. The Kerala judgment is the Act’s only full merits vindication, and its limits should be noted with the same care as its holding: it was a decision of a High Court on maintainability and broad validity, delivered while the Supreme Court’s own examination stood unresolved, and it decided nothing about Section 17(b) or Section 46 as between operator and supplier, because no supplier was before it and no incident had occurred.

Fukushima, Kudankulam and the Act’s first years

The Act commenced on 11 November 2011, eight months after the earthquake and tsunami of March 2011 destroyed the Fukushima Daiichi plant in Japan, and the accident reordered the setting in which the new statute had to operate. A law drafted in the afterglow of the 2008 diplomatic opening took effect in a world where nuclear safety politics had hardened everywhere, where the costs of a severe accident were being counted in public in yen, and where every figure in the Indian Act could be compared, unfavourably and instantly, with the compensation demands of an actual event. The comparison did the cap no favours: within five years Japanese compensation and decontamination spending had passed sums that dwarfed India’s overall statutory maximum, and the critics of Section 6 acquired an exhibit that no projection could have supplied.

At home the Act’s first years were dominated by Kudankulam, where the Russian reactors, contracted long before the Act, met a protest movement of fishing communities and farmers organised under the People’s Movement Against Nuclear Energy, led by S.P. Udayakumar, which blockaded the site through 2011 and 2012. The movement’s demands ran to the plant’s safety and to the liability framework alike, and the litigation it generated reached the Supreme Court in G. Sundarrajan v Union of India, where a bench of the Court in May 2013 declined to halt the plant’s commissioning, issued directions on safety and oversight, and treated the existence of the liability statute as part of the framework within which the project could proceed. The judgment matters to this article for a distinction it illustrates. Sundarrajan was a safety and siting case; it was not a decision on the validity of the liability Act’s cap, still less on Section 17, and its references to the compensation framework were part of a survey of the regulatory landscape rather than a holding on any contested provision. Campaigners and governments both cited it for more than it decided, a habit this article has tried not to repeat.

The Fukushima shadow also lengthened the negotiation the Act governed. India’s civil nuclear agreement with Japan, signed in November 2016 and brought into force in July 2017, mattered to the supplier question because Japanese heavy industry stood behind parts of the American and French supply chains, Toshiba then owning Westinghouse and Japanese forges supplying reactor pressure vessels for the designs under negotiation. Liability assurances formed part of that negotiation’s background as they did of every other. By the middle of the decade, therefore, the Act faced in every direction at once: foreign vendors declining its recourse clause, domestic protesters citing its cap as proof of the programme’s priorities, two constitutional challenges unresolved or confined to one High Court, and a government committed in public to the statute’s conformity with international norms while constructing, piece by piece, the apparatus of clarification, rule and pool whose only purpose was to make the statute’s central clause bearable to the parties it governed. That is the posture in which the fifteen years began.

Fifteen years without a contract

The empirical record of the Act is a list of things that did not happen, set against a list of things that did, and the discipline this article has applied to negative claims should be observed: the record is stated with attribution, project by project.

Why did Russia build reactors in India while American suppliers held back?

Because the Russian units at Kudankulam rested on intergovernmental agreements concluded in 1988 and 1998, before the Act existed, and were built by a state vendor answerable to its own government rather than to a board pricing litigation risk. American vendors, answerable to shareholders and insurers, treated the statutory exposure as a term they could not sign against.

Kudankulam in Tamil Nadu is the exception that identifies the rule. Its first two Russian VVER reactors were contracted under intergovernmental agreements that predated the Act by more than a decade, their construction was far advanced when the Act commenced, and the Russian state vendor, building as an instrument of its government under agreements negotiated government to government, proceeded through the Act’s lifetime, recording reservations about the liability law’s application to later units while continuing to build. Russian supply to India was never, in structure, a commercial response to the Act; it was a state commitment that the Act arrived too late to govern and that Russian diplomacy managed around the statute rather than through it.

The Russian programme also supplies the one instance of the Act’s machinery being negotiated around rather than merely complained about. Units 3 and 4 at Kudankulam proceeded under a general framework agreement concluded in 2014, after the Act’s commencement, with construction following in the Act’s later years and further units agreed in principle behind them. The liability allocation for those units was managed within the framework the Act and the insurance arrangements supplied, which is why the Russian track is best described not as exempt from the Act but as absorbed by a relationship in which the vendor’s government could carry, politically, risks that a commercial board could not carry legally. The contrast with the American and French tracks is therefore not between a vendor who accepted the Act and vendors who rejected it; it is between a state that could underwrite its vendor’s discomfort and markets that would not.

The American projects show the Act’s operation in its intended field. The 2008 agreements allotted sites for American reactors, and the site at Kovvada in Andhra Pradesh was designated for six reactors, with Westinghouse’s AP1000 the technology in negotiation after an earlier designation involving GE Hitachi designs. The Nuclear Power Corporation and Westinghouse concluded preliminary arrangements, including an early works agreement in 2016, and the two governments repeatedly announced that a commercial contract was the object of continuing negotiation. Westinghouse entered bankruptcy protection in the United States in March 2017 under the weight of its American construction projects, which scrambled the negotiation afresh; the company emerged under new ownership, the talks continued, and no construction or supply contract was concluded. GE Hitachi, for its part, had made the vendor position plainest in public: its leadership indicated that it would not expose the company to the Indian liability regime as enacted, and no GE reactor project advanced beyond designation.

The French project completes the pattern. Jaitapur in Maharashtra, allotted for six European Pressurised Reactors, began with a general framework agreement between the Nuclear Power Corporation and Areva in December 2010, months after the Act’s assent; Areva’s nuclear business passed to EDF, and in April 2021 EDF submitted what it described as a binding techno-commercial offer for six EPR units. An offer is not a contract. The liability allocation remained among the questions the negotiation had to settle, alongside tariff and financing, and construction did not begin. The wider programme within which these projects sit, its sites, its targets and its stalled expansions, is mapped in the guide to infrastructure laws in India; the point for this article is narrower. Every foreign project the Act was passed to enable remained, at the Act’s repeal, at or before the stage of offers and preliminary agreements, and in the two cases where a vendor’s reasoning is documented in public, liability stood among the stated causes. Other causes were real, and honesty requires listing them: the tariffs Indian distribution companies could pay for nuclear power against falling renewable prices, the financing cost of first-of-a-kind reactors, land and protest politics at the sites, and in Westinghouse’s case a bankruptcy unconnected with India. The verdict below weighs liability among those causes rather than above them; what cannot be sustained is the proposition that liability played no part, because the vendors, their governments and the Indian government’s own five-year settlement effort all behaved as if it did.

The replacement: the SHANTI Act of 2025

Parliament closed the argument in December 2025 by repealing the statute it was about. The Sustainable Harnessing and Advancement of Nuclear Energy for Transforming India Act, 2025, the SHANTI Act (Act 39 of 2025), was introduced in the Lok Sabha on 15 December 2025 by Minister Jitendra Singh, passed by the Lok Sabha on 17 December and by the Rajya Sabha on 18 December, received the President’s assent on 20 December 2025, and came into force on 21 December 2025. It repeals both the Atomic Energy Act, 1962 and the Civil Liability for Nuclear Damage Act, 2010, and replaces them with a single framework that opens nuclear generation to private and joint venture participation under licence, gives the safety regulator a statutory footing, and rewrites the liability chapter. The 2010 settlement this article has adjudicated therefore closed at the end of 2025, not by judicial decision and not by the slow victory of one side’s legal argument, but by repeal: the framework within which the recourse controversy was fought ceased to exist, and the questions the controversy raised about the earlier statute survive only as the history of a repealed law and as the starting point from which the replacement’s different choices have to be understood.

On the question this article has followed, the replacement is a concession written into law. The statutory right of recourse against a supplier for defective supply, the clause (b) that the 2010 Parliament inserted and two governments defended, does not survive: supplier liability under the new framework flows through the commercial contract between operator and supplier, which is the first limb of the old Section 17 standing alone, and the operator’s liability is graded by installation size with caps rising to three thousand crore rupees for the largest facilities, backed by a statutory nuclear liability fund. The fifteen-year executive settlement, clarification, rule, pool and ratification, had attempted to reach by interpretation the position the 2025 Act reaches by enactment: recourse as a matter of bargain rather than of statute. There is an irony in the replacement’s parentage that deserves a sentence, because it is the kind of irony that settles arguments in democracies. The Bharatiya Janata Party, which in 2010 attacked the Bill for being written to foreign specifications and demanded stronger supplier exposure, is the party whose government in 2025 enacted the statute that deletes the recourse clause, caps the operator’s liability by grade, and presents the whole as the framework that will finally unlock supply. Political positions on liability have tracked the difference between opposition and government more closely than they have tracked any theory of liability, and the clause’s defenders and its deleters have been, in part, the same political movement at different distances from responsibility for the programme’s outcomes. That does not make either position insincere. It makes the episode legible as politics: the clause was the price of carrying the Bill in 2010, and its deletion was the price of building reactors in 2025, and the price was paid by the parties in whose interest the programme stood at each date.

Two cautions belong with the description. The replacement is recent, its subordinate legislation has been the subject of public consultation by the Department of Atomic Energy rather than long operation, and no incident, claim or supply contract has tested its liability chapter; it is described here as enacted, not as proven. And its arrival does not retrospectively answer the argument of principle that produced clause (b): the question whether a supplier of defective equipment should answer for the defect has been settled by Parliament in the supplier’s favour as a matter of policy for the future, not refuted as a matter of reasoning about the past.

The replacement has itself attracted the mirror image of the old litigation. Petitions challenging the SHANTI framework have attacked the features this article’s second critique attacked in the 2010 Act, the capping of liability and the removal of supplier exposure, and in hearings on the new law the Supreme Court has clarified that the statutory framework does not restrict the constitutional courts’ authority to determine fair and just compensation in an appropriate case. That clarification, offered from the bench while the petitions stand unresolved, identifies the permanent tension the 2010 Parliament was also navigating and which no drafting has dissolved: a liability statute can cap what its own machinery will pay, and it cannot cap what a constitutional court, faced with an actual catastrophe and actual victims, may hold the right to life requires. The suppliers of 2010 understood that proposition from the other side, which is why a saving clause and a statutory scheme never exhausted their analysis of their exposure. The episode’s deepest lesson sits there. In a constitutional democracy with an activist higher judiciary, channelling is always provisional, and every party to the nuclear liability debate, vendor, government and victims’ movement alike, was in truth negotiating over how much provisional certainty each would accept.

Eight things said about the Act that the record does not support

The claim that the Act prohibited supplier liability is the reverse of the truth. The Act’s contested feature was that it preserved and created supplier exposure: a statutory recourse in Section 17(b) and an unlatched door in Section 46. Accounts that describe the 2010 Act as a shield for suppliers are describing the international model, or the 2025 replacement, not the statute that governed for fourteen years.

The claim that the 2015 clarifications amended the Act is wrong in law. The Ministry of External Affairs’ answers of 8 February 2015 stated the government’s interpretation; they altered no word of the statute, bound no court, and were understood by every lawyer who commented on them, for the government and against it, as interpretation rather than amendment. The five years of negotiation they were meant to close continued after their publication, which is the practical proof of their legal weight.

The claim that Rule 24 capped all recourse under Section 17 overstates the rule. Rule 24 of the 2011 Rules prescribed the content of contracts under clause (a): a recourse measure set at the lower of the operator’s statutory liability and the contract value, running for the initial licence period or the product liability period, whichever is longer. Its application to a clause (b) claim arising independently of contract was contested from the start, and a former Attorney General opined that the rule was vulnerable as inconsistent with the Act. Treating the rule as a complete answer repeats the government’s negotiating position as if it were a holding.

The claim that the operator’s cap was fifteen hundred crore rupees for every installation misstates Section 6. The figure applies to nuclear reactors of ten megawatts thermal power or more. Spent fuel reprocessing plants carry a cap of three hundred crore rupees, and research reactors below that threshold, other fuel cycle facilities and transport carry one hundred crore. The overall maximum for a single incident, the figure the Central Government’s tier runs up to, is the rupee equivalent of three hundred million Special Drawing Rights.

The claim that the courts struck the Act down, or upheld it in the Supreme Court, are both wrong. The Supreme Court admitted a petition for examination in 2012 and delivered no concluding judgment on validity during the Act’s life. The reasoned merits decision is the Kerala High Court’s dismissal of a public interest challenge, which upheld the enactment against the Part III attack made on it. Anything stronger, in either direction, is invention.

The claim that the insurance pool was part of the 2010 Act confuses the statute with the settlement built around it. The Act obliged the operator to obtain insurance or other financial security under Section 8, and said nothing about a pool. The Indian Nuclear Insurance Pool was an executive creation of June 2015, five years after enactment, assembled by GIC Re and eleven other insurers as one limb of the package meant to make the Act workable. The distinction matters because the pool could be, and was, designed around the Act’s difficulties, while amending the Act itself was the step no government of the period was willing to take in Parliament.

The claim that the Convention on Supplementary Compensation required India to delete clause (b) overstates what any authority decided. Supplier governments argued that the clause was inconsistent with the Convention’s annex, India argued that its law conformed, and the Convention provides no tribunal to decide between them. No ruling, award or authoritative opinion ever held the Indian Act non-conforming. The clause’s fate was decided by commerce and then by Parliament, not by international adjudication, and accounts that present the deletion as a legal compulsion misdescribe a political and commercial outcome as a judicial one.

The claim that India stood outside the international liability system until 2025 is wrong. India signed the Convention on Supplementary Compensation in October 2010 and ratified it on 4 February 2016, with the Convention in force for India from 4 May 2016. The controversy was never whether India would join the system; it was whether one clause of its national law conformed to the system’s central rule, argued by the Indian state in one sense and by the system’s other members in another.

The propositions the record supports

Before the verdict, the fixed points, because the public debate on this Act decays fastest when its anchors slip. The Act is Act 38 of 2010, assented to on 21 September 2010, brought into force on 11 November 2011. Its operator is the Central Government or its authority, corporation or Government company licensed under the Atomic Energy Act, which in the power sector meant the Nuclear Power Corporation of India Limited, and that operator is strictly liable under Section 4 subject to the Section 5 exclusions for grave natural disaster of exceptional character, armed conflict, hostility, civil war, insurrection and terrorism. The overall maximum liability for one incident is the rupee equivalent of three hundred million Special Drawing Rights under Section 6(1); the operator’s share is fifteen hundred crore rupees for a power reactor, three hundred crore for a spent fuel reprocessing plant, and one hundred crore for research reactors below ten megawatts thermal, other fuel cycle facilities and transport, under Section 6(2); the Central Government supplies the balance under Section 7, may build a Nuclear Liability Fund by levies, and did so from December 2015. The operator must insure under Section 8, and the Indian Nuclear Insurance Pool of fifteen hundred crore rupees, launched by GIC Re and eleven other insurers on 12 June 2015, became the instrument of that insurance, offering also a suppliers’ right of recourse policy. Claims lie to a Claims Commissioner or, where established, a Nuclear Damage Claims Commission, with civil courts barred by Section 35, and rights are extinguished after ten years for property and twenty years for personal injury under Section 18. Recourse lies under Section 17 on three grounds, of which the second, supply of equipment or material with patent or latent defects or sub-standard services, is found in no counterpart convention text. Section 46 saves other law and names the operator alone. The 2011 Rules bounded contractual recourse by reference in Rule 24; the Ministry of External Affairs stated the government’s reading on 8 February 2015; India signed the Convention on Supplementary Compensation on 27 October 2010 and ratified it on 4 February 2016, the Convention taking effect for India on 4 May 2016. The Supreme Court agreed in March 2012 to examine the Act’s validity and delivered no concluding judgment on the question in the Act’s life; the Kerala High Court dismissed a public interest challenge on the merits it heard. No foreign reactor supply contract was concluded under the Act. The Act was repealed and replaced by the SHANTI Act, Act 39 of 2025, assented to on 20 December 2025 and in force from 21 December 2025, which removes the statutory supplier recourse, confines supplier liability to the contract, recalibrates the operator’s liability by installation grade, and places the safety regulator on a statutory footing. A reader who holds those propositions holds the controversy; every other statement in the debate is commentary on one of them.

One discipline in reading the list deserves emphasis, because it is the discipline this article has tried to observe throughout. Almost every proposition in it is a proposition about conduct rather than about intention: about what the section says rather than what its authors hoped, about what the vendors did rather than what they professed, about what the governments built around the statute rather than what they said about it. The controversy lasted fifteen years largely because its participants kept exchanging statements of intention, conformity, principle and reassurance, while the operative facts sat in a short statutory clause, a negatived amendment, an insurance market’s refusal and a file of unsigned contracts. Statutes in this field are not finally interpreted by the commentaries written about them. They are interpreted by the behaviour they succeed or fail in organising, and the behaviour is the record.

The verdict

The stance charter of this series requires a verdict where the evidence supports one, and on the Act’s operation the evidence is one-directional in a way it rarely is.

The Act as operated deterred supply; whether it would have delivered compensation was never tested. No nuclear incident occurred in India during the Act’s life, so its compensation machinery, commissioner, fund and tiers were never run, and any claim that they would have paid victims promptly and adequately is as unprovable as any claim that they would have failed. What the Act demonstrably did, in the world of contracts and boards where its effects can be observed, is stand for fifteen years as a stated reason, by named vendors and named governments, why reactor supply contracts could not be signed, and consume five years of executive effort in clarifications, rules, insurance engineering and treaty ratification whose sole purpose was to neutralise one of its clauses without amending it. When the neutralisation was finally accomplished, it was accomplished by repeal and re-enactment on the model’s terms. A statute whose own government spent half its life working around it, and whose replacement deletes its signature provision, has been judged by the only tribunal whose judgment on a commercial statute finally matters: the sustained behaviour of everyone required to act under it.

That verdict convicts the drafting strategy, not the principle that animated it. The parliamentarians who insisted on clause (b) were answering a real question, and the answer they rejected, a public operator indemnifying a foreign supplier against the supplier’s own defective equipment out of public funds, is an answer a self-respecting legislature could refuse and many electorates would have punished it for accepting. Their error was instrumental. A statutory recourse keyed to latent defect, uncapped in the section and unbounded in time, does not transfer risk to the party best able to control it; it transfers the project to a different country, because risk that cannot be bounded cannot be insured, and risk that cannot be insured cannot be signed for by a board with duties to anyone else. The instruments that might have reconciled the principle with the market were all visible in 2010: recourse confined to the contract but with the contract’s mandatory minimum terms prescribed by statute, as Rule 24 later attempted for clause (a) alone; a defect limb with a stated ceiling and a stated sunset written into the Act itself rather than into a vulnerable rule; and a Section 46 that said in terms what the government always insisted it meant. Each was available to the drafters of Section 17 and Section 46. None was taken, because taking any of them would have required conceding in the text what the politics of the passage required be asserted in the text: that the supplier stood exposed.

The costs of the assertion are measurable in the programme that did not happen: three allotted foreign projects unbuilt across a decade and a half in which India’s electricity demand and its climate commitments both pressed for firm low-carbon capacity, and a domestic programme carried entirely by indigenous designs that the foreign track was meant to supplement. The benefits are harder to state because they are counterfactual: no Indian victim was ever denied recovery against a supplier under the Act, because there was no incident; and no supplier ever paid a rupee under clause (b), because there was no contract from which a recourse claim could grow. A provision that operates only as a deterrent, and deters the very transactions its parent statute exists to enable, has failed on its own terms even if the principle inside it was sound. The SHANTI Act’s quiet deletion of the clause, fifteen years on, by a Parliament of a different political colour from the one that inserted it, and with the argument conducted almost entirely in the language of investment rather than of Bhopal, is the record’s own conclusion. The clause was inserted deliberately, its effect was foreseen and argued about before enactment, and the foresight of its critics, as prediction, was borne out in full. What the episode establishes for every statute in this series that follows is the namable claim with which this article began: a liability Act is read most carefully not by its commentators but by the risk committees of the parties it governs, and one limb of one section, if it makes a risk unpriceable, is enough to stop a programme.

Two objections to that verdict should be faced rather than left to the reader. The first is that the verdict proves too much from a negative: reactor projects fail for many reasons, land, tariffs, financing, the collapse of a vendor’s home business in Westinghouse’s case, and the counterfactual in which a compliant liability Act produces six French and six American reactors on schedule is a fantasy no serious analyst would sign. The objection is sound against a verdict of sole causation, and this article’s verdict is not that. It is that liability was a stated, documented and independently sufficient obstacle, attested by the vendors, by two allied governments and by the Indian government’s own five-year campaign against its own clause, and that a statute cannot be acquitted of an outcome merely because other causes were also at work in the same field. The second objection runs the other way: that the verdict is too kind to the clause’s defenders, because a principle asserted at the cost of a programme, against advice, and abandoned by the asserters’ own political heirs, deserves a harder word than instrumental error. The answer is that the principle has not in fact been abandoned as a principle; it has been relocated. The SHANTI framework leaves supplier liability to contract, which means it leaves to the Nuclear Power Corporation and its successors the task of extracting from each vendor, in negotiation, the defect responsibility that Parliament in 2010 tried to impose by law. Whether the Indian state proves a better negotiator of recourse than it was a legislator of it is the live question the new Act hands to the next fifteen years, and it will be answered in contract documents the public will never read, which is itself a measure of what was given up when the argument moved out of the statute book.

There is a last symmetry to record. The Act was passed to complete a framework for victims and became famous for what it did to suppliers; it was defended for fifteen years as victim-protective and repealed in the name of investment; and the one constituency that never obtained a forum under it, the victims of an incident, is the constituency whose absence from the entire story is the story’s luckiest fact. India argued for a decade and a half, at the highest levels of three governments, over who would pay for a disaster that did not happen. The statute that argument produced is gone, replaced by one whose liability chapter is untested for the same happy reason. If the replacement is ever tested, the country will learn in the worst circumstances what its capped, channelled, contractual scheme is worth. The 2010 Act’s epitaph is that its worth was assessed, exhaustively and at leisure, by everyone except events.

Frequently Asked Questions

Q: What did the Civil Liability for Nuclear Damage Act provide?

It created a complete compensation code for nuclear incidents in India. Liability was channelled to the operator of the installation and made strict, so a victim needed to prove the incident and the damage rather than anyone’s fault. Liability was capped, the operator was required to carry insurance or financial security, claims were routed to a Claims Commissioner or a Nuclear Damage Claims Commission instead of the civil courts, and rights were extinguished after fixed periods. The Act, numbered 38 of 2010, received assent on 21 September 2010 and came into force on 11 November 2011.

Q: What was the operator’s liability cap?

Section 6(2) graded the cap by installation. For nuclear reactors with thermal power equal to or above ten megawatts, the operator’s maximum liability for one incident was Rs 1,500 crore. For spent fuel reprocessing plants it was Rs 300 crore, and for research reactors below ten megawatts, other fuel cycle facilities and the transport of nuclear materials it was Rs 100 crore. Above the operator’s cap, the Central Government answered for the balance up to the overall maximum for a single incident, fixed by Section 6(1) at the rupee equivalent of 300 million Special Drawing Rights.

Q: When can an operator exercise the supplier right of recourse?

Section 17 gave the operator a right of recourse after paying compensation in three situations. The right could be expressly provided for in a contract in writing. It arose where the nuclear incident resulted as a consequence of an act of the supplier or the supplier’s employee, which includes the supply of equipment or material with patent or latent defects or sub-standard services. And it arose where the incident resulted from an act of commission or omission of an individual done with the intent to cause nuclear damage. The second ground, independent of any contract, was the contested departure from the international model.

Q: Why did suppliers object to the liability law?

Suppliers objected chiefly to clause (b) of Section 17 and to Section 46. Clause (b) exposed them to recourse whenever an incident could be traced to equipment or material with a latent defect, a standard that could be applied decades after supply, with no ceiling or time limit stated in the section itself. That exposure could not be priced or insured on ordinary terms. Section 46, which kept the Act in addition to other law and preserved proceedings against the operator, left suppliers fearing direct suits by victims outside the Act, unconstrained by its cap, its forum and its limitation periods.

Q: What did Section 46 mean for suppliers?

Section 46 provided that the Act was in addition to, and not in derogation of, any other law in force, and that nothing in the Act exempted the operator from proceedings that might otherwise be instituted against the operator. It named the operator and not the supplier. The Government of India maintained that the clause therefore preserved no claim against suppliers and that amendments to write suppliers into it had been moved in Parliament and negatived. Suppliers answered that the clause preserved other law generally and never stated that no proceeding could lie against a supplier, so a direct tort claim outside the Act remained arguable.

Q: How was a nuclear damage claim made?

The person sustaining injury, the owner of damaged property, the legal representatives of a deceased person, or an authorised agent applied to the Claims Commissioner appointed for the area of the incident, after the Atomic Energy Regulatory Board had notified the incident. The Commissioner inquired by a summary procedure and made an award. Where the Central Government considered the damage sufficiently widespread, a Nuclear Damage Claims Commission performed the function instead. Section 35 barred civil courts from matters within the Commissioner or the Commission’s remit. Rights were extinguished after ten years for property damage and twenty years for personal injury under Section 18.

Q: What was the nuclear insurance pool?

The Indian Nuclear Insurance Pool was launched on 12 June 2015 by the General Insurance Corporation of India, GIC Re, with eleven other Indian insurers, with a capacity of Rs 1,500 crore, matching the operator’s statutory cap. It offered the operator’s liability policy required by Section 8 of the Act and a suppliers’ right of recourse policy under which suppliers could insure against the exposure Section 17 created. The New India Assurance Company issued the policies on the pool’s behalf. The pool made the operator’s security obligation performable domestically, although it could not insure a supplier against a direct claim brought outside the Act.

Q: What has replaced the nuclear liability law?

The Sustainable Harnessing and Advancement of Nuclear Energy for Transforming India Act, 2025, the SHANTI Act (Act 39 of 2025), assented to on 20 December 2025, repealed the Civil Liability for Nuclear Damage Act, 2010 together with the Atomic Energy Act, 1962 and replaced both with a single framework. The new law opens nuclear generation to private and joint venture participation, grades the operator’s liability cap by installation size up to Rs 3,000 crore for the largest facilities, and confines supplier liability to what the supply contract provides, removing the statutory recourse for defective supply that the 2010 Act had created.

Q: What does Section 17(b) of the nuclear liability Act say?

Clause (b) of Section 17 gave the operator a right of recourse where the nuclear incident has resulted as a consequence of an act of the supplier or his employee, which includes supply of equipment or material with patent or latent defects or sub-standard services. Three features made it the centre of the controversy. It applied whether or not the contract mentioned recourse. It was triggered by defect rather than by intent or by a pleaded case of negligence in ordinary form. And the section stated no ceiling on the recovery and no period after supply within which the exposure ended.

Q: How much could the Central Government have to pay after a nuclear incident?

The Central Government stood behind the operator’s cap. Under Section 7, where claims exceeded the operator’s maximum, the government was liable for the difference up to the overall maximum for one incident, the rupee equivalent of 300 million Special Drawing Rights, a figure the Department of Atomic Energy’s own answers valued at about Rs 3,167 crore at the rate cited. Beyond that ceiling the Act contemplated the government taking additional measures, and after 2016 the international fund under the Convention on Supplementary Compensation stood behind the national tier. The government was also empowered to build a Nuclear Liability Fund from levies on operators.

Q: How long did a victim have to claim compensation under the Act?

Section 18 extinguished the right to claim if the claim was not made within ten years from the date of the nuclear incident for damage to property, and within twenty years from the date of the incident for personal injury to any person. The periods ran from the incident, not from the discovery of the harm. Critics attacked that design for radiation injury, since a disease with a long latency could declare itself after the period had closed, leaving the injured person outside the Act however diligent they had been, and however strong the medical evidence connecting the illness to the incident.

Q: Why did the Government issue clarifications on the liability Act in 2015?

The clarifications, published by the Ministry of External Affairs as frequently asked questions and answers on 8 February 2015, were issued to unblock reactor negotiations with supplier countries, above all the United States, whose administration had pressed the liability question at government level. They stated the government’s interpretation of Section 17 as permissive rather than mandatory for the operator, confined contractual recourse to the contract’s own limits, denied that Section 46 created supplier liability, and asserted that Indian law conformed to the Convention on Supplementary Compensation. They formed one limb of a settlement package completed by the insurance pool and by ratification of the Convention.

Q: Can a set of official answers change what an enacted statute means?

No. A clarification issued by the executive states the government’s interpretation of a statute; it does not amend the statute, and a court deciding a claim is free to read the enacted words differently. That objection, made by commentators when the 2015 answers appeared, went to the heart of the settlement’s weakness: Section 17(b) conferred a right in statutory terms, and an answer sheet saying the operator was free not to exercise it could not bind a future operator, a future government or a court. The clarification’s force was diplomatic. It recorded the Indian state’s position for foreign governments, while vendors’ lawyers continued to advise on the text.

Q: Why was the compensation cap criticised as too low?

Critics compared the statutory maxima with the demonstrated cost of industrial and nuclear disasters. The Bhopal gas disaster produced a settlement of 470 million United States dollars in 1989 for a chemical plant, with medical consequences argued over for decades. The Fukushima accident generated compensation, decontamination and decommissioning costs exceeding India’s overall maximum of 300 million Special Drawing Rights by orders of magnitude. Against those measures, an operator cap of Rs 1,500 crore and an overall ceiling fixed in 2010, never revised although Section 6 allowed revision by notification, looked like figures chosen for insurability and treaty conformity rather than for the scale of the harm insured against.

Q: What did Rule 24 of the 2011 Rules do to the right of recourse?

Rule 24 of the Civil Liability for Nuclear Damage Rules, 2011 prescribed the content of contracts carrying recourse under clause (a) of Section 17. The contract had to include a right of recourse measured by the lower of the operator’s liability under Section 6(2) and the value of the contract, and the recourse was to run for the duration of the initial licence under the Atomic Energy (Radiation Protection) Rules, 2004, or the product liability period undertaken in the contract, whichever was longer. Suppliers and their advisers noted that the rule addressed clause (a) contracts and did not in terms bound a claim under clause (b).

Q: Did the courts strike down the nuclear liability Act?

No court struck it down. The Supreme Court admitted a public interest petition by Common Cause for examination of the Act’s validity in March 2012, a bench headed by Chief Justice S.H. Kapadia observing that the Act had a link to the right to life, but no concluding judgment on validity followed during the Act’s life. The reasoned merits decision came from the Kerala High Court, where a Division Bench of Chief Justice Ashok Bhushan and Justice A.M. Shaffique dismissed a public interest challenge to a long list of provisions, finding no infirmity, arbitrariness or violation of fundamental rights in the enactment.

Q: What is the Convention on Supplementary Compensation and when did India join it?

The Convention on Supplementary Compensation for Nuclear Damage, adopted in 1997 under the International Atomic Energy Agency, establishes a worldwide liability regime built on channelling liability to the operator and adds an international fund, financed by contributions from member states, above the national compensation tier. India signed the Convention on 27 October 2010, a month after the liability Act received assent, deposited its instrument of ratification on 4 February 2016, and the Convention entered into force for India on 4 May 2016. The Indian government presented ratification as confirmation that its law conformed to the Convention; supplier governments continued to test the claim against Section 17(b).

Q: Why did the operator being a public undertaking matter to the recourse debate?

Because channelling works differently where the operator is the State. In the markets that produced the international model, a commercial operator buys the plant, negotiates the supply terms and prices liability risk into its business, so exclusive channelling places the cost of a supplier’s defect on a party able to bargain it away in advance. In India the operator was the Nuclear Power Corporation, a government company, insured through a largely public pool, with the tier above its cap payable by the Central Government. Exclusive channelling would therefore have made the Indian taxpayer the insurer of a foreign supplier’s defective equipment, which is the proposition Parliament refused in 2010.

Q: Which projects were meant to bring foreign reactors to India under the Act?

Three sites carried the foreign programme. Kudankulam in Tamil Nadu hosted Russian VVER reactors under intergovernmental agreements from 1988 and 1998 and proceeded through the Act’s lifetime. Jaitapur in Maharashtra was allotted for six French European Pressurised Reactors, beginning with a framework agreement with Areva in 2010 and reaching a binding techno-commercial offer from EDF in April 2021. Kovvada in Andhra Pradesh was designated for American reactors, with Westinghouse’s AP1000 in negotiation and an early works agreement signed in 2016. Neither the French nor the American project reached a construction or supply contract during the Act’s life.

Q: How does the SHANTI Act change the liability of suppliers?

The SHANTI Act, 2025 removes the statutory right of recourse that Section 17(b) of the 2010 Act had given the operator where an incident resulted from equipment or material with patent or latent defects or sub-standard services. Under the replacement framework, a supplier’s liability for nuclear damage flows through the commercial contract it signs with the operator, which is the position the international model prescribes and the first limb of the old Section 17 standing alone. The change was the central liability concession of the 2025 legislation and the formal end of the fifteen-year dispute this article records.